Characterizing environmental factors contributing to plant community dynamics on a full-scale green roof in Philadelphia, PA

Background/Question/Methods

Much of the research done on green roofs considers them from either a horticultural standpoint or from an engineering perspective. However, green roofs are becoming more commonly implemented in urban landscapes in order to combat lost ecosystem services in these areas, and thus, deserve study from an ecological viewpoint as well. Our study site was a green roof located in Philadelphia, PA which was designed with four different drainage layer test plots. It was initially planted in 2010 with a uniform mix of forbs, grasses and Sedum species, but plant coverage and distribution changes have occurred.  Our objective was to assess how four environmental parameters (wind, shade, temperature, and substrate water content) along with drainage layer treatment influence plant community dynamics of this full-scale roof.  Detailed plant coverage data was gathered at 24 locations across the roof, and these coverage results were correlated with growing season temperature range, maximum, and minimum, substrate water content, and shade duration at each location. Wind measurements were taken at nine locations using a laser Doppler anemometer. Physiological measurements (gas exchange and water potential) were also used to further understand plant distribution and abundance patterns on the roof.

Results/Conclusions

Our results indicated the total cover was mostly influenced by temperature range and substrate water content; cover was higher in locations with more moderate temperature extremes and higher water content. Wind speed, turbulence, and shade duration did not show significant effects on overall plant coverage. Drainage layer treatment also showed no significant effects on total cover or substrate water content by plot. Analysis by functional group showed that Sedum cover was also greatly influenced by temperature range and substrate water content, and forb cover was influenced by temperature minimum and water content. Individual plant species coverage was not uniform across the roof. In some cases we were able to test hypotheses regarding differences in coverage.  For example, physiological measurements indicated that the difference in abundance between Bouteloua curtipendula and Liatris spicata is due to differences in photosynthesis rates while having similar transpiration rates and drought tolerances of the species. We also considered differences at larger roof scales; all patterns shown at the sampling location scale explain larger scale patterns seen across the roof.  Our study shows that green roof plant communities are responding to small scale habitat differences which create an ecologically dynamic green roof system.